基于改进BP网络的人脸信息获取算法

人脸识别技术在各领域都有着广泛的应用前景,但人脸信息量过于巨大,就目前的计算机处理能力而言,尚不能完美地解决这个问题,但通过一定的技术达成人脸信息识别的部分功能还是可行的.该文利用BP算法,分别设计了适合于完成部分脸信息提取分类的神经网络结构,测试了几种训练算法在人脸信息识别应用中的实际效果,并通过性能比较选出了适用的学习算法,最终在人脸是否带墨镜、表情、性别识别上达到较好的效果.

An Algorithm of Human Face Identification Based on Rapid BP Neural Network

The human face identifying technology has a wide and promising prospect in many fields. However the complexity of this problem is beyond the capability of computers today. In spite of this, it is still possible to implement parts of the face identifying function. In this paper, Rapid BP Neural Network technology is employed to acquire some information of face, some special Neural Network structures are designed to implement some special face identification functions. The actual effect of several training algorithms is tested in the face identification fields and the best one is selected after careful comparison. Eventualy the design gets a satisfactory result on glasses, facial expression and sex identification.